Field surface roughness levelling of the lapping metal surface using specular white light

Abstract

For robotic machining, accurate and automatic inspection of finished surface is necessary for implementation in the site lapping process. Modern inspection systems based on smart sensor technology such as image processing and machine vision have been widely spread into many industries. These systems along with the smart factory concept not only enhance the inspection accuracy but also decrease human works substantially. In this paper, we propose a method for automatic levelling of machined surface with respect to roughness values, adopting specular light-based vision technique. The study mainly concerns the development of surface roughness levelling system associated with textural analysis related to surface topography. It is supported by the fundamental property of light reflection: reflection changes from diffuse to specular depending upon surface texture. A rough surface having tool marks produces contrast in grayscale values, resulting in the decrease of intensity value and vice versa. Image processing technique was adopted to find the underlying grayscale values of inspected surface. The result showed a nonlinear increase in grayscale values as roughness decreases. The highest image resolution can be achieved when surface normal corresponds to perspective center of camera, so the concept was extended for inclined and curved surfaces. To obtain high accuracy in precise measurements, a multiscale measuring method was developed for a wide range of roughness, which does not require an isolated system, but only change in camera distance for high-resolution measurement. The proposed technique showed surface roughness levelling with high accuracy and resolution up to 20 nm (R_a). The results indicate that this technique can be used for multiscale surface levelling of the free-form metal surfaces.